INSTITUTE OF PHYSICS PUBLISHING EUROPEAN JOURNAL OF PHYSICS

Eur. J. Phys. 28 (2007) 183190 doi:10.1088/0143-0807/28/2/004

Lorentz transformations with arbitrary

line of motion
Chandru Iyer1 and G M Prabhu2
1 Techink Industries, C-42, Phase-II, Noida, India
2 Department of Computer Science, Iowa State University, Ames, IA, USA

E-mail: prabhu@cs.iastate.edu

Received 18 October 2006

Published 2 January 2007
Online at stacks.iop.org/EJP/28/183

Abstract
Sometimes it becomes a matter of natural choice for an observer (A) that he
prefers a coordinate system of two-dimensional spatial xy coordinates from
which he observes another observer (B) who is moving at a uniform speed
along a line of motion, which is not collinear with As chosen x- or y-axis. It
becomes necessary in such cases to develop Lorentz transformations where the
line of motion is not aligned with either the x- or the y-axis. In this paper we
develop these transformations and show that under such transformations, two
orthogonal systems (in their respective frames) appear non-orthogonal to each
other. We also illustrate the usefulness of the transformation by applying it to
three problems including the rod-slot problem.

1. Introduction

In most treatments on special relativity [13], the line of motion is aligned with the x-axis.
This is a natural choice because in such a situation the (y, z) coordinates are invariant under
the Lorentz transformations (y
= y and z
= z). However, it is of interest to study the case
when the line of motion does not coincide with any of the coordinate axes. Practical instances
of such a situation are an airplane during landing or take off. The ground at the airfield has a
natural coordinate system with the x-axis parallel to the ground, whereas the airplane ascends
or descends at an angle with the ground. In this paper we develop such a formulation for
a two-dimensional spatial system which can also be extended to three-dimensional spatial
systems.
The pedagogical contribution of this paper lies in the derivation of the Lorentz
transformation when the line of motion is not aligned with any of the axes. In two-dimensional
space the algebra is manageable as demonstrated by its application to the rod-slot problem.
Since the algebra becomes more complex in three-dimensional space, the student will realize
the advantages of aligning one of the coordinate axes with the line of motion and appreciate
why this is the accepted practice.
0143-0807/07/020183+08$30.00

c 2007 IOP Publishing Ltd Printed in the UK 183
184 C Iyer and G M Prabhu

y y y y

K
K

x

x

K K

x x
v

Figure 1. Relative motion of K (on left) and K
(on right).

2. Development of generalized 2D Lorentz transformations

The transformation matrix for planar rotation by an angle in the anti-clockwise direction is
given by


cos sin
.
sin cos
So
x
= x cos + y sin y
= x sin + y cos .
Within an inertial frame, when we add the time coordinate and consider the fact that a spatial
rotation has no effect on the time coordinate, we add the equation t
= t, thus getting the
transformation for spatial rotation to be


x cos sin 0 x
y
= sin cos 0 y . (1)
t
0 0 1 t
The last row of the matrix indicates t
= t, and the two zeros in the last column of the matrix
indicates there is no cross effect of t on x
or y
.
We use the symbol R to denote this transformation, indicating spatial rotation anti-
clockwise by an angle .
Similarly, the Lorentz transformation along the x-axis yields the following matrix in a
two-dimensional spatial system:


x 0 v x
y
= 0 1 0 y , (2)
t
v /c2 0 t
where = 1
.
1v 2 /c2
The middle row of the matrix, (0, 1, 0), indicates y
= y, and the two zeros in the middle
column of the matrix indicate that there is no cross effect of y on x
or t
. We use the symbol
Lxv to denote a Lorentz transformation of magnitude v along the x-axis.
When the line of motion is inclined at an angle with the x-axis, we can use the
transformation R() LxvR between two frames K and K
which observe each other moving at
a velocity of +v and v, respectively as shown in figure 1. As observed by both the frames,
Lorentz transformations with arbitrary line of motion 185

the proper angle of the line of motion is with respect to their respective x-axes. Noting that
cos( ) = cos and sin( ) = sin , we obtain the matrix A for R() Lxv R :

cos2 + sin2 sin . cos ( 1) v cos
A = sin cos ( 1) sin2 + cos2 v sin . (3)
v cos v sin
c 2 c 2
The above matrix A transforms the event coordinates (x, y, t) of frame K to (x
, y
, t
) of frame
K
when K and K
are moving at a proper angle (anti-clockwise with respect to each others
positive x, x
-axis respectively) and the relative velocity of K
with respect to K is +v.
The inverse of matrix A, denoted as B, is obtained by substituting v for v:

cos2 + sin2 sin cos ( 1) v cos
B = sin cos ( 1) sin2 + cos2 v sin . (4)
v cos v sin
c2 c2

The matrix B transforms the event coordinates (x
, y
, t
) of frame K
to (x, y, t) of frame K,
when K and K
are moving at a proper angle (anti-clockwise with respect to each others
positive x, x
-axis respectively) and the relative velocity of K with respect to K
is v. By
using standard trigonometric identities, it can be verified that AB = BA = I, the identity
matrix.



x x
Thus if we designate the vector y to be K and the vector y

to be K
, we have
t t

K
= AK and K = BK
.
To visualize the x
-axis as seen by K we let y
= 0 and t = 0 (this is an instant in K, all points
with y
= 0). For this condition to hold we obtain from K
= AK the following equation:
a21 x + a22 y = 0. (5)
The slope of this line is a21 /a22 = sin cos (1 )/[ sin2 + cos2 ].
One can verify that this line has a negative slope and is not aligned with the x-axis; this
is notwithstanding the fact that the proper angle between the line of motion and the x-axis is
the same in K and K
. Further we can infer (from the fact that this slope is not dependent on
the sign of v) that the slope of the x
-axis as observed from K is the same as the slope of the
x-axis as observed from K
.
Similarly the slope of the y
and y axes as observed from K and K
are also negative.
Figure 2 depicts the manner in which two orthogonal systems appear non-orthogonal to each
other. In general, the quadrants of the moving frame through which the line of motion (drawn
through the origin) passes, expand; the other two quadrants of the moving frame contract, as
observed by an observer co-moving with the stationary
frame.

3. Application of the transformation to an object moving at an incline with

respect to a rod

Consider a rod of proper length L. Let us say an object is moving at a velocity v and at an
angle with the axis of the rod (as observed from the co-moving inertial frame of the rod).
What will be the apparent length of the rod as observed by an observer co-moving with the
inertial frame of the object?
We have a preferred coordinate system (F) co-moving with the rod as shown in figure 3.
In this system the x-axis is along the length of the rod and one end of the rod has coordinates
x = 0; y = 0. The other end of the rod has coordinates x = L; y = 0.
186 C Iyer and G M Prabhu

y y
y y

v
v

v 90
90
x x

x x
K as observed by K v K as observed by K

Figure 2. Observations from K (on left) and K
(on right): orthogonal proper systems appear
non-orthogonal.

y
y O
x

Object moving
at velocity v
F
x

(0, 0) Rod (L, 0)

Figure 3. Rod in frame F and object in frame O moving at velocity v.

The spacetime coordinates of the two end points of the rod at any arbitrary time t read (0,
0, t) and (L, 0, t). We can also visualize a coordinate system (O) co-moving with the object
with the x
-axis aligned with the line of relative motion.
In order to transform from F to O we use the transformation [LxvR ] as specified in
equations (1) and (2). The resultant transformation is

0 v cos sin 0
T = Lxv R = 0 1 0 sin cos 0
v /c 0 2
0 0 1

cos sin v
= sin cos 0 . (6)
v cos v sin
c2 c2

We can apply matrix T as derived in equation (6) to transform the event coordinates from F to
O. One end of the rod at t = 0 in F has the coordinates (0, 0, 0) and it transforms to (0, 0, 0)
in O. This corresponds to t
= 0. In order to observe the other end of the rod from the inertial
Lorentz transformations with arbitrary line of motion 187

y y

Stationary Rod
K
x

K
x
Moving Rod

Figure 4. Rod in frame K
moves towards stationary rod in frame K at velocity v.

frame O at t
= 0, we transform the coordinates of the other end of the rod at some instant t in
frame F and set t
= 0.


x L
y
= T 0 . (7)
0 t
From equation (7) we obtain

x
= L cos v t (8)


y = L sin (9)
0 = (Lv cos )/c2 + t. (10)

Solving equation (10) for t and substituting its value in equation (8) gives x
= L cos / .
Thus at the instant t
= 0 in O, one end of the rod has coordinates x
= 0; y
= 0, and the
other end has coordinates x
= L cos

/ and y = L sin . Therefore the apparent length
of the rod as observed by O is L (cos / ) + sin2 . This quantity is equal to L when
2 2

= 1 (no relative motion), is equal to L/ when = 0 (traditional Lorentz transformation

along line of motion) and is equal to L when = 90 (no variation in length perpendicular
to line of motion). All these results are consistent with known concepts of Lorentzian
contraction.

4. Application of the transformation to the collision of inclined rods problem

Another suitable problem for application is the collision of the inclined rods as described in
[4]. Here we have two parallel (considering proper angles) rods in relative motion with the
axis of both the rods making a proper angle with the line of relative motion. If both the
rods select their respective axis as their x- and x
-axis we can use the transformations A and B
in equations (3) and (4) as the forward and inverse transformations between the two inertial
frames.
In figure 4, the bottom collision is taken as the spacetime origin (0, 0, 0) in both the frames
K and K
[4]. The top collision has the coordinates (L, 0, t) in frame K and (L, 0, t
) in frame
K
. Using matrix A from equation (3) to transform (L, 0, t) and equating the result to (L, 0, t
)
188 C Iyer and G M Prabhu

a
Rod

CR
b
CS
vx
W
Center of rod CR approaching center of slot CS
along the line of motion W v = (vx 2 + vy 2 )
vy

Figure 5. Initial conditions of the rod-slot problem.

we obtain the following equations.

( cos2 + sin2 )L (v cos )t = L (11)

L sin cos ( 1) (v sin )t = 0 (12)
(v /c2 )L cos + t = t
. (13)

From equations (11) and (12) we get the same result, namely t = (L/v) cos [1 (1/ )].
This redundancy is due to our prior assumption that the top end of the first rod collides with the
top end of second rod. Substituting this into equation (13) yields t
= (L/v) cos [(1/ ) 1].
This is consistent with the results in [4] and confirms the reversal in time order of the top and
bottom collisions in the two inertial frames co-moving with each of the two rods respectively.

5. Application of the transformation to the rod and slot problem

Another suitable application for the transformation derived above is the rod and slot problem
described in [5]. Here the proper angles are unequal and this presents a more general
application. In this scenario the rod exhibits motion in two directions, but only with constant
velocity. Furthermore, the line of motion (that is, the line joining the centres of the rod and
slot) is not aligned with either the axis of the rod or the slot, there is no gravity, and thus no
stress or propagation of stress.
We assume the proper angle of the axis of the rod with the line of motion to be  and
that of the slot to be . The rod has a coordinate system with the x-axis along the axis of the
rod and the origin at the centre of the rod. Similarly the slot has a coordinate system with
the x-axis along the axis of the slot and the origin at the centre of the slot. The meeting of
the centres of the rod and slot is the spacetime origin with (x, y, t) = (x
, y
, t
) = (0, 0, 0) as
shown in figure 5. The rod is shown parallel to the slot in figure 5 ( = ), but in general 
does not have to be equal to .
When we transform from the rods reference frame to the slots reference frame, we have
to use the following transformations in sequence. R() will be the first transformation to align
the line of motion with the x-axis. Then we do a Lorentz transformation of magnitude v
to switch to the slots reference frame; this is designated as Lx(v). The resultant coordinates
are in the slots reference frame but with the x-axis aligned with the line of motion. Since the
slot has a preferred system with the x-axis aligned with the slots axis, we need to perform
Lorentz transformations with arbitrary line of motion 189

a transformation R(). Thus the overall transformation is R()Lx(v)R(). Denoting this

transformation as D, we obtain the elements of D as follows:

cos cos + sin sin sin cos + sin cos v cos
D = cos sin + cos sin sin sin + cos cos v sin . (14)
v cos v sin
c2 c2

Let a denote half the length of the rod, and b denote half the length of the slot. The leading
edge of the rod has the coordinates x = a; y = 0 for any arbitrary time t in the rods frame. This
can be denoted by the triple (a, 0, t). Similarly the front edge of the slot has the coordinate
(b, 0, t
) in the slot
s frame for any arbitrary time t
in the slots frame. In order for the rod to
just pass through the slot, (a, 0, t) must transform to (b, 0, t
) for some value of t.

b a
Thus 0 = D 0 .
t
t
This gives rise to the following three equations:

a cos  cos + a sin sin  + vt cos = b (15)

a cos  sin + acos sin  vt sin = 0 (16)
(av /c2 ) cos  + t = t
. (17)

From equation (16) we get

t = (a cos sin  a cos  sin )/(v sin ).
Substituting for t in equation (15) we get
a cos  cos + a sin sin  + (a cos sin  a cos  sin ) cos / sin = b.
Multiplying both sides by sin and simplifying, we obtain
a sin (sin2 + cos2 ) = b sin
or
a sin  = b sin .
Thus we find that in order for the rod to just pass through the slot, we obtain a condition that
depends only on the four proper quantities, that is, the proper lengths of the rod and slot and
the proper angles between the line of motion and the axis of the rod and slot. Further, it is
easy to deduce that whenever a sin  < b sin , the rod passes through the slot.
It can be further deduced from the above discussions that if the line of motion passing
through the centre of the rod does not intersect the slot at its centre, but divides the slot
into two unequal lengths b1 and b2, then the condition for the rod to pass through the slot is
given by
(a sin  < b1 sin ) AND (a sin  < b2 sin ).

6. Summary

We derived a general Lorentz transformation in two-dimensional space with an arbitrary line of

motion. We applied it to three problems and demonstrated that it leads to the same solution as
already established in the literature. The solved problems using the transformation equations
190 C Iyer and G M Prabhu

illustrate the convenience of the chosen coordinate system in either frame of reference. Thus,
in certain situations, we see the merit of using the Lorentz transformations with the line of
motion not coinciding with any of the coordinate axes. Our method can also be conceptually
extended to three-dimensional space with an arbitrary line of motion. However, the algebra
is a lot more complex in this case. This may be one reason that conventionally the Lorentz
transformations are performed with the line of motion coinciding with one of the coordinate
axes.

References
[1] Einstein A 1961 Relativity, The Special and General Theory (New York: Three Rivers Press) (Authorized
translation by Rober W Lawson)
[2] Taylor E and Wheeler J 1992 Spacetime Physics (San Francisco: Freeman) p 70
[3] Resnick R 1968 Introduction to Special Relativity (New York: Wiley)
[4] Iyer C and Prabhu G 2006 Reversal in the time order of interactive events: the collision of inclined rods Eur. J.
Phys. 27 81924
[5] Iyer C and Prabhu G 2006 Differing observations on the landing of the rod into the slot Am. J. Phys. 74 9981001